Beruflich Dokumente
Kultur Dokumente
a r t i c l e
i n f o
Article history:
Received 16 October 2014
Received in revised form 6 July 2015
Accepted 7 July 2015
Keywords:
Anxiety
Depression
Insulin therapy
Insulin-naive people
Type 2 diabetes
Prospective study
a b s t r a c t
Background: Depression and anxiety have been found to be predictors of poor health outcomes in diabetes, but
mechanisms are still unclear.
Aims: To examine whether symptoms of anxiety and depression were associated with timing of initiating insulin
therapy.
Methods: A cohort study of insulin-naive particpants with type 2 dabetes completed the Hospital Anxiey and Depression Scale, HADS-A (n = 731) and/or the HADS-D (n = 768) in the communy-based Nord-Trndelag Health
Study (1995-1997). Information on insulin initiation was retrieved from the Norwegian Prescription Database
from January 1, 2004 to November 21, 2012. Cox regression analyses were used to estimate the association between symptoms of anxiety, depression and time to insulin initiation.
Results: At baseline, 19% reported anxiety symptoms (score 8) and 18% depressive symptoms (score 8). After a
mean follow-up of 4.4 (SD 3.6) years, 337 (40%) participants had started insulin therapy. After adjustment for
sociodemographic and clinical variables, anxiety symptoms were associated with later initiation of insulin therapy (HR 0.70, 95% CI 0.490.99), while depressive symptoms were not. Considering groups simultaneously, having both elevated depressive and elevated anxiety symptoms was associated with later time to insulin initiation
(HR 0.62, 95% CI 0.390.99), while having only anxiety symptoms (without depressive) HR 0.81, 95% CI
0.501.32) or only depressive symptoms (without anxiety) (HR 1.08, 95% CI 0.681.72) were not.
Conclusions: Anxiety was associated with a later initiation of insulin, while depressive symptoms were not.
Persons with both elevated levels of anxiety and depression were also less likely to start insulin therapy. These
results need further testing in other prospective studies.
2015 Elsevier Inc. All rights reserved.
Introduction
People with type 2 diabetes mellitus have an increased risk of anxiety and depression [14]. Though meta-analyses and longitudinal studies have shown that anxiety and depression are associated with
suboptimal glycemic control, a higher risk of diabetes complications
and higher mortality rates among people with diabetes [511], it is
not fully understood why emotional problems are associated with
poorer subsequent health outcomes. One mechanism might be the
http://dx.doi.org/10.1016/j.jpsychores.2015.07.004
0022-3999/ 2015 Elsevier Inc. All rights reserved.
timing of insulin therapy. In the long term, insulin therapy is unavoidable for the majority of people with type 2 diabetes, and yet many individuals are reluctant to start insulin [12]. In the United Kingdom
Prospective Diabetes Study, for example, 27% of participants randomized to insulin therapy refused insulin, while the refusal rate of oral
medication was 7% in the glibenclamide group and 13% in the chlorpropamide group [13]. Anxiety and/or depression may contribute to a
delay in insulin therapy. People with anxiety or depression often have
cognitive difculties, low self-esteem and concentration problems, suffer from fatigue and do not have enough energy to take on new tasks,
such as starting insulin therapy [14]. Cross-sectional studies have
found depression to be associated with a more negative appraisal of insulin therapy among insulin-naive people with type 2 diabetes, which
may underlie a reluctance to start insulin treatment [1416].
Little information from longitudinal studies is currently available on
whether anxiety and/or depressive symptoms are associated with later
310
Methods
HUNT2, which was conducted during 19951997 has been described in detail previously [18,19]. In brief, HUNT2 was a communitybased survey including residents of Nord-Trndelag County aged 20
years. Of all people invited for the study, 70% attended and returned
the general self-report questionnaire which included information on
prevalent diabetes. Participants who responded afrmatively to the
question Do you have or have you had diabetes? were classied as
having diabetes (n = 1972) [20]. During a follow-up appointment for
those who responded positively to this question (75% participated), a
fasting blood sample was drawn to determine the type of diabetes.
Participants were classied as having type 1 or type 2 diabetes based
on a combination of three factors: (1) measures of fasting C-peptide
and glutamic acid decarboxylase antibodies (anti-GAD), (2) fasting glucose levels and (3) for those taking insulin, time from diagnosis to
starting insulin treatment [18]. Participants were also asked about diabetes treatment. Those who answered that they did not use insulin
were classied as respondents with insulin-naive type 2 diabetes
(n = 998). We used the NorPD to verify that participants who reported
diabetes but did not answer the treatment question did not use insulin
(n = 126). The follow-up sample included 1124 participants with selfreported diabetes who were not treated with insulin (Fig. 1).
Fig. 1. Study owchart for the cohort recruited from HUNT2 (19951997) and linked to
data from the Norwegian Prescription Database (NorPD). Footnote: aSelf-reported diabetes, veried as type 2 diabetes (n = 783); only self-reported diabetes (n = 215); self-reported diabetes, not answering treatment question but veried as not using insulin in the
NorPD (n = 126).
Statistical analysis
We tabulated descriptive statistics with mean (standard deviation
(SD)) and proportions for baseline characteristics of the total sample.
We used Student's t-tests and Pearson chi-square tests or if not appropriate Fisher exact tests to compare the characteristics of participants
with HADS-A or HADS-D scores b 8 with those scoring 8.
Survival analyses were used to investigate the time until the rst
prescription of insulin treatment during the follow-up period. The survival time was from the day the NorPD started (January 1, 2004) until
the rst registered insulin prescription. We censored the survival
times of those with no prescriptions at the date of death or at study closure at November 21, 2012. Survival curves were estimated by the
KaplanMeier method for participants who scored 8 and for those
who scored b8 on the HADS-A and HADS-D. Log rank tests were used
to compare survival distributions. The proportional hazards assumption
was checked by visual inspection of the KaplanMeier survival curves
(using Cox regression with a time-dependent covariate only when
these suggested violation) and the Harrell & Lee test based on the
Schoenfeld residuals.
Cox regression analyses were used to estimate the hazard ratios
(HRs) of insulin initiation along with 95% condence intervals (CIs).
Initially, univariable Cox regression analyses were performed for all
baseline covariates. For covariates with N10% missing data, separate
unknown categories were used. This involved education (n = 111)
and any microvascular complication (n = 191). We then performed hierarchical multivariable Cox regression analyses with adjustment for
other known factors associated with insulin initiation. Covariates were
grouped thematically and increasingly complex models were developed
by adding one set of variables at a time using forced entry (models 13).
We rst adjusted the analyses for age and sex (model not shown).
Then the following blocks of factors were included: low level of education (no/yes) and living alone (no/yes) (model 1). Model 2 also included severity of illness (judged by HbA1c level) (continuous), and model 3
included any cardiovascular disease (no/yes), any microvascular
complications (no/yes) and time since diagnosis of type 2 diabetes.
Analyses were also performed using HADS-A and HADS-D as a continuous scale. An additional analysis was also performed using a combined
311
Table 1
Baseline characteristics for the study sample of participants with type 2 diabetes not using insulin, by anxiety symptoms and depressive symptoms (HADS-A and HADS-D scores of b8
versus 8).a
Characteristics
n = 839
Sociodemographic characteristics
Age, mean, (SD), years
65 (12)
0
Men, % (n)
47 (393)
0
Living alone,% (n)
37 (312)
3
Low education, % (n)
65 (470) 111
Clinical characteristics
HbA1c,mean (SD), mmol/mol (IFCC
61 (19)
45
units)
7.7 (1.7)
45
HbA1c, mean (SD), % units (NGSP
units)
Time since diabetes diagnosis
6.7 (7.6)
35
6 (48)
31
Stroke, % (n)c
Myocardial infarction, % (n)c
10 (80)
21
Angina pectoris, % (n)c
16 (131) 24
d
Any cardiovascular disease, % (n)
24 (199) 22
Microalbuminuria, % (n)
19 (142) 98
Eye problems caused by diabetes, %
8 (50) 185
(n)c
Any microvascular complication, % 28 (181) 191
(n)d
a
HADS-A symptom
scores 8 (n = 136)
HADS-D symptom
P-valueb
scores 8 (n = 135)
64 (12)
50 (296)
38 (223)
62 (337)
62 (13)
41 (56)
32 (43)
67 (83)
0.10
0.07
0.19
0.28
64 (12)
48 (305)
38 (237)
62 (355)
65 (11)
44 (60)
30 (40)
72 (84)
0.67
0.43
0.08
0.04
62 (19)
56 (15)
0.001
61 (19)
58 (14)
0.22
7.8 (1.7)
7.3 (1.4)
0.001
7.7 (1.7)
7.5 (1.3)
0.22
6.7 (7.7)
5 (27)
9 (54)
15 (84)
22 (128)
19 (103)
6 (27)
7.4 (8.7)
6 (8)
11 (14)
19 (26)
24 (32)
18 (22)
13 (14)
0.39
0.55
0.66
0.17
0.66
0.71
0.005
6.6 (7.1)
5 (32)
9 (55)
14 (88)
22 (136)
19 (109)
6 (31)
7.7 (10.6)
8 (10)
14 (18)
24 (32)
31 (42)
16 (20)
12 (12)
0.26
0.27
0.09
0.005
0.02
0.46
0.03
26 (124)
32 (33)
0.24
26 (132)
31 (30)
0.32
The total population varies somewhat depending on the actual completion of the different tests and questionnaires (missing range 0191).
P-value for comparing the respective groups: HADS-A b 8 versus HADS-A 8 and HADS-D b 8 versus HADS-D 8.
Stroke, myocardial infarction, angina pectoris and eye problems due to diabetes are self-reported.
d
All medical comorbid conditions were combined into two composite disease measures: any cardiovascular disease (myocardial infarction, stroke and/or angina pectoris) and any
microvascular complications (eye problems due to diabetes and/or microalbuminuria).
b
c
312
The rate of insulin initiation did not differ signicantly between people
with and without elevated symptoms of depression (35% versus 42%,
P = 0.11). The KaplanMeier curves for the groups with anxiety scores
8 versus b8 (Fig. 2a) differed signicantly (P = 0.008): people with
anxiety symptoms (score 8) had a longer mean time to insulin initiation than those with a symptom score b 8 (6.5 years, 95% CI 5.97.1 versus 5.5 years, 95% CI 5.25.8). The KaplanMeier curves did not differ for
groups with depressive symptom scores 8 versus symptom scores b 8
(Fig. 2b; P = 0.16), with a mean time to insulin initiation of 6.1 (95% CI
5.56.8) and 5.6 (95% CI 5.36.0) years, respectively. Median survival
times could not be reported, since less than half of the total sample
started insulin in the follow-up period. Neither the KaplanMeier
plots nor the correlation between the Schoenfeld residuals and ranked
survival time (r = 0.02, P = 0.74 for HADS-A and r = 0.05, P = 0.43
for HADS-D) suggested that the proportional hazards assumption had
been violated.
Bivariate Cox regression analysis (n = 656) showed that people
with a high level of anxiety (symptom scores 8) were less likely to
start insulin therapy during the whole follow-up period than their counterparts with scores b 8 (unadjusted HR = 0.64, 95% CI, 0.450.89; P =
0.01). In multivariate analyses (n = 656), simultaneous adjustment for
sex and age did not alter the association between anxiety and a later
start of insulin therapy, nor did introducing marital status and educational level (model 1, Table 2). However, further adjustment for HbA1c
(model 2, Table 2) resulted in less strong effect, i.e. a hazard ratio closer
to 1 that was no longer statistically signicant (HR = 0.72; 95% CI
0.511.02; P = 0.06). Subsequent adjustment for any cardiovascular
disease (myocardial infarction, stroke and/or angina pectoris), any microvascular complication (eye problems due to diabetes and/or
microalbuminuria) and time since diagnosis of diabetes resulted in a
statistically signicant difference (HR = 0.70, 95% CI 0.49-0.99; P =
0.045). In fully adjusted analyses with HADS-A as a continuous variable,
the relation with time to insulin initiation was non-signicant (HR =
0.99, 95% CI 0.951.02; P = 0.42) (Table 3). There was no statistically
signicant departure from linear trend (P = 0.06).
Depressive symptoms (8) were not signicantly associated with
time to insulin initiation (n = 689; HR = 0.81, 95% CI 0.591.126;
P = 0.21). The association with insulin initiation remained nonsignicant after adjustment for sex and age (P = 0.24). Subsequent adjustment for living alone status and educational level (P = 0.24), HbA1c
level (P = 0.28) and vascular complications and time since diabetes diagnosis (P = 0.29) did not change the association either (right panel,
Table 2). This was also the case with HADS-D as a continuous variable
in fully adjusted analyses (HR 1.02, 95% CI 0.981.05; P = 0.36)
Fig. 2. KaplanMeier curves for the rst prescription of insulin treatment comparing individuals with type 2 diabetes with HADS-A and HADS-D symptom scores 8 versus b8.
313
Table 2
Univariable and multivariable Cox proportional hazards models for the association between baseline anxiety symptoms (HADS-A scores) and insulin initiation (left side of table) and
between baseline depressive symptoms (HADS-D scores) and insulin initiation (right side of table).a.
HADS-Anxiety
score b 8
HADS-Anxiety
score 8
HADS-Anxiety scores
continuous
HADS-Depression
score b 8
HADS-Depression
score 8
HADS-Depression
score continuous
Age (years)
Men
Low educatione
Living alone
HbA1c
Cardiovascular diseasef
Microvascular
complicationse,g
Time since diabetes
diagnosis
Unadjusted HR
(CI)
Model 1 HR
(CI)
Model 2 HR
(CI)
Model 3 HR
(CI)
Unadjusted HR
(CI)
Model 4 HR
(CI)
Model 5 HR
(CI)
Model 6 HR
(CI)
n = 656b
n = 656
n = 656
n = 656
n = 689c
n = 689
n = 689
n = 709
Reference
Reference
Reference
Reference
0.64 (0.450.89)
0.63 (0.450.88)
0.72 (0.511.02)
0.70 (0.49-0.99)
0.98 (0.941.01)
0.97 (0.941.00)
0.99 (0.961.02)
0.99 (0.95-1.02)
Referenced
Referenced
Referenced
Referenced
0.81 (0.591.12)
0.82 (0.601.14)
0.84 (0.611.16)
0.84 (0.601.17)
1.00 (0.961.03)
1.00 (0.961.03)
1.01 (0.981.05)
1.02 (0.981.05)
0.98 (0.970.99)
1.63 (1.282.06)
0.80 (0.621.01)
0.80 (0.621.04)
1.35 (1.271.43)
0.96 (0.711.28)
1.39 (1.051.84)
0.98 (0.970.99)
1.50 (1.121.92)
1.04 (0.791.36)
0.89 (0.691.15)
0.98 (0.970.99)
1.42 (1.111.82)
0.90 (0.691.18)
0.77 (0.591.00)
1.38 (1.301.46)
0.98 (0.970.99)
1.39 (1.071.78)
0.90 (0.691.19)
0.78 (0.601.01)
1.36 (1.281.45)
1.02 (0.751.38)
1.21 (0.891.63)
0.98 (0.970.99)
1.60 (1.262.01)
0.82 (0.641.05)
0.83 (0.651.06)
1.36 (1.281.43)
0.93 (0.701.23)
1.36 (1.031.80)
0.98 (0.970.99)
1.52 (1.201.93)
1.09 (0.841.43)
0.92 (0.711.18)
0.98 (0.970.99)
1.39 (1.091.77)
0.93 (0.711.21)
0.79 (0.611.02)
1.38 (1.301.46)
0.98 (0.970.99)
1.36 (1.061.75)
0.93 (0.711.22)
0.81 (0.621.05)
1.38 (1.301.46)
1.00 (0.741.35)
1.15 (0.861.54)
1.01 (0.991.03)
1.01 (0.991.03)
1.01 (0.991.03)
1.01 (0.991.02)
Data are hazard ratios (HR) with 95% CI. HR for the covariates was estimated in a model with HADS-A or HADS-D categorized (8 versus b8).
n = 673 distributed as n = 127 reporting HADS anxiety symptoms 8 versus n = 546 b 8.
c
n = 709 distributed as n = 128 reporting HADS depressive symptoms 8 versus n = 581 b 8.
d
Reference category.
e
Missing values entered as a separate category.
f
Self-reported angina, stroke and/or myocardial infarction, as reported at baseline.
g
Albumin/creatinine ratio N3.0 mg/mmol in at least two of the three urine samples was used to dene microalbuminuria. Eye problems caused by diabetes and/or microalbuminuria
were combined into a composite measure of microvascular complications.
b
most infections are of short duration and often will be handled without
profound changes in diabetes treatment. Fourth, we do not know
whether people not using insulin or with later insulin initiation had
been offered insulin treatment previously by their physician. Fifth, the
source population is ethnically homogeneous (3% were of nonCaucasian origin), and the results therefore may not be generalized to
other ethnic populations. Finally, HADS was designed as a screening
measure and not as a diagnostic tool [28]. The ndings based on the continuous HADS scores were only partly in line with results for dichotomized HADS scores. On the one hand, dichotomizing the continuous
HADS scores might be questioned from a methodological point of
view (e.g. articial cut off, loss of statistical power). On the other
hand, tests for linearity showed a departure for linearity for HADSDepression, indicating that these analyses need to be interpreted with
Table 3
Univariable and multivariable Cox proportional hazards models for the association between HADS-A and HADS-D combined and insulin initiation: baseline low score on HADS-A and
HADS-D (reference group) versus those with 1) elevated anxiety only, 2) elevated depression only and 3) elevated anxiety and depression scores combined.a
Unadjusted HR (CI)
Model 1 HR (CI)
Model 2 HR (CI)
Model 3 HR (CI)
n = 652b
n = 652
n = 652
n = 652
Referencec
0.63 (0.401.01)
0.95 (0.601.50)
0.63 (0.400.99)
0.98 (0.970.99)
1.67 (1.322.11)
0.78 (0.610.99)
0.79 (0.621.02)
1.35 (1.271.43)
0.94 (0.711.26)
1.38 (1.041.81)
1.01 (1.001.03)
Referencec
0.63 (0.391.01)
0.98 (0.621.55)
0.62 (0.390.99)
0.98 (0.970.99)
1.49 (1.161.90)
1.05 (0.801.38)
0.90 (0.701.17)
Referencec
0.85 (0.531.37)
1.07 (0.671.69)
0.64 (0.411.02)
0.98 (0.970.99)
1.42 (1.101.82)
0.91 (0.691.19)
0.78 (0.601.01)
1.38 (1.301.46)
Referencec
0.81 (0.501.32)
1.08 (0.681.72)
0.62 (0.390.99)
0.98 (0.970.99)
1.38 (1.071.78)
0.91 (0.691.20)
0.79 (0.601.03)
1.36 (1.281.45)
1.01 (0.741.38)
1.22 (0.901.65)
1.01 (0.991.03)
314
caution. For anxiety the same test was almost signicant, which might
be due to power. Therefore the ndings for HADS continuous need to
be interpreted with caution. It has also been debated whether the dichotomizing of the HADS is valid across languages and culture [29].
However, the cut-off level applied in this study has been validated for
a similar Norwegian population [22].
Forty percent of the participants had started insulin therapy after a
mean follow-up of 4.4 (SD 3.6) years. This is consistent with previous
research [30], although the SD was larger in our study (29 months versus 43 months in our study). At baseline, respondents with an increased
level of anxiety (8) appeared to have more optimal HbA1c levels than
those with scores b 8. The fact that participants with a high level of anxiety had lower HbA1c levels may indicate that anxiety is associated with
a healthier lifestyle. Perhaps anxious people strive to live healthier lives
in order to avoid the long-term complications of diabetes. Previous research has shown a relation between personality and adherence to diabetes treatments. For example, Lane et al. [31] reported that persons
with type 2 diabetes who had higher scores on neuroticism had lower
HbA1c levels ( 0.22). Anger hostility ( 0.23) and vulnerability
( 0.25) were also associated with lower HbA1c levels. In contrast,
higher altruism scores were also associated with poorer glycemic control (0.41). However, in a more recent study [32], individuals with diabetes who had a Type D (distressed) personality less often reported to
follow a healthy diet or to consult healthcare professionals in case of
problems with diabetes management. After adjustment for differences
in baseline HbA1c, the HR related to anxiety symptoms became less
strong, changing from 0.64 to 0.72, and became statistically nonsignicant (P = 0.06). A later start of insulin therapy in people with anxiety symptoms may partly be due to more optimal glucose control in
that particular group. In this complicated picture anxiety may therefore
have a protective effect in type 2 diabetes. However, even though the association between anxiety symptoms and insulin initiation was no longer signicant after controlling for HbA1c, the hazards ratio was still
0.72. This may provide credit for the existence of alternative explanations for a later start of insulin therapy in people who are anxious,
such as psychological insulin resistance. Future studies should take
into account that in the present study, the combination of anxietyand depressive symptoms was associated with a later start of insulin
treatment, while depressive symptoms only (without having anxiety
symptoms) and anxiety symptoms only (without having depressive
symptoms) were not associated. Nefs et al. [17] did not nd an association between depressive symptoms and a later start of insulin initiation,
but they did not take anxiety into account. In case the present ndings
are replicated, future studies should also explore the mechanisms. It is
conceivable that this particular group fears insulin injections, or lack
self-esteem and worry about the complexity of insulin therapy. They
may also worry about insulin-related weight gain, or about the increased risk of serious hypoglycemic events that are associated with insulin therapy. Concerns about negative reactions from others might also
play a role, or the feeling that insulin initiation indicates the patient's
failure to manage diabetes with oral medication. Further studies need
to investigate the potential health-compromising and healthprotective role of anxiety and its associations with initiation of insulin
therapy and glycemic control.
In line with our ndings, people with type 2 diabetes in the DAWN
study reported substantial worry about starting insulin [33]. One
might speculate that anxious people with type 2 diabetes more often
fear insulin injections or blood glucose self-testing, which has shown
to be a barrier to insulin treatment in previous studies [3436]. If a
later initiation of insulin treatment in people with anxiety symptoms
represents an inappropriate delay, providers may need to address the
specic symptoms of anxiety that might negatively affect appraisals of
insulin therapy [35]. Strong negative appraisals of insulin therapy
have been reported among almost one third of insulin-naive people
[12], and depressive symptoms have been associated with negative appraisals regarding insulin [1416]. In the current study, participants
with elevated depressive symptom scores (8) were 20% less likely to
start insulin treatment than those with a low level of depressive symptoms (HADS-D b 8). However the difference was non-signicant. It
seems that anxiety symptoms play a more important role in the timing
of insulin initiation than depressive symptoms. Further investigations
should assess whether the appraisal of insulin therapy (Insulin Treatment Appraisal Scale scores) [37] of people with type 2 diabetes is
more strongly associated with anxiety than depression. We found
later initiation of insulin therapy in men compared to women, and
higher age associated with later initiation of insulin therapy. The existence of different health beliefs across gender and age groups might
be a reason for later initiation of insulin therapy. For example, previous
research showed that women have more psychological barriers to insulin treatment compared to men [35]. Our ndings regarding age and sex
differences in timing of insulin therapy warrant further investigation in
other studies.
Previous work on the health care provider's perspective emphasized
the apparent complexity of the medication choice process, which seems
to contrast with current evidence-based treatment guidelines [36],
since guidelines emphasize the need for early addition of insulin treatment to meet treatment goals [25]. The DAWN study reported that
most nurses and general practitioners (5055%) had a tendency to
delay insulin therapy until absolutely necessary [33]. Further, there
might be cultural differences in the initiation of insulin therapy; the
delay has been reported to be lower among providers in Scandinavia
than in the United States [33]. However, we do not have information
on the role of health care providers in dealing with anxiety and/or depressive symptoms in assessing the timing of insulin initiation.
In conclusion, in this longitudinal study we found no association between elevated depressive symptoms and a later insulin initiation,
while anxiety symptoms were associated with insulin initiation. We
also found that people who reported both elevated anxiety and depressive symptoms were less likely to start insulin therapy than those with
low scores, while depressive symptoms only (without having anxiety
symptoms) and anxiety symptoms only (without having depressive
symptoms) were not associated. These results contribute to strengthen
the rationale for identifying high-risk patients with mental health
symptoms and to tailor their needs.
Conict of interest statement
The authors declare that they have no competing interests.
Acknowledgments
The Nord-Trndelag Health Study (The HUNT Study) is a collaboration between HUNT Research Centre (Faculty of Medicine, Norwegian
University of Science and Technology NTNU), Nord-Trndelag County
Council, Central Norway Health Authority, and the Norwegian Institute
of Public Health.
References
[1] A Nouwen, K Winkley, J Twisk, CE Lloyd, M Peyrot, et al., Type 2 diabetes mellitus as
a risk factor for the onset of depression: a systematic review and meta-analysis,
Diabetologia 53 (2010) 24802486.
[2] AB Grigsby, RJ Anderson, KE Freedland, RE Clouse, PJ Lustman, Prevalence of anxiety
in adults with diabetes: a systematic review, J. Psychosom. Res. 53 (2002)
10531060.
[3] KJ Smith, M Beland, M Clyde, G Gariepy, V Page, et al., Association of diabetes with
anxiety: a systematic review and meta-analysis, J. Psychosom. Res. 74 (2013) 8999.
[4] B Mezuk, WW Eaton, S Albrecht, SH Golden, Depression and type 2 diabetes over
the lifespan: a meta-analysis, Diabetes Care 31 (2008) 23832390.
[5] RJ Anderson, AB Grigsby, KE Freedland, M de Groot, JB McGill, et al., Anxiety and
poor glycemic control: a meta-analytic review of the literature, Int. J. Psychiatry
Med. 32 (2002) 235247.
[6] FE van Dooren, G Nefs, MT Schram, FR Verhey, J Denollet, et al., Depression and risk
of mortality in people with diabetes mellitus: a systematic review and metaanalysis, PLoS One 8 (2013) e57058.
315
[22] I Bjelland, AA Dahl, TT Haug, D Neckelmann, The validity of the Hospital Anxiety and
Depression Scale. An updated literature review, J. Psychosom. Res. 52 (2002) 6977.
[23] A Mykletun, E Stordal, AA Dahl, Hospital Anxiety and Depression (HAD) scale: factor
structure, item analyses and internal consistency in a large population, Br. J. Psychiatry 79 (2001) 540544.
[24] M Kjaergaard, CE Arfwedson Wang, K Waterloo, R Jorde, A study of the psychometric properties of the Beck Depression Inventory-II, the Montgomery and sberg Depression Rating Scale, and the Hospital Anxiety and Depression Scale in a sample
from a healthy population, Scand. J. Psychol. (Nov 21 2013)http://dx.doi.org/10.
1111/sjop.12090.
[25] American Diabetes Association, Standards of medical care in diabetes2013, Diabetes Care 36 (2013) S11S66.
[26] TJ Hastie, RJ Tibshirani, Generalized Additive Models, Chapman & Hall, London,
1990.
[27] MS Kirkman, VJ Briscoe, N Clark, H Florez, LB Haas, et al., Diabetes in older adults,
Diabetes Care 35 (2012) 26502664.
[28] RP Snaith, The Hospital Anxiety and Depression Scale, Health Qual. Life Outcomes I
(2003) 29.
[29] GA Maters, R Sanderman, AY Kim, JC Coyne, Problems in cross-cultural use of the
hospital anxiety and depression scale: no butteries in the desert, PLoS One 9
(2013) e70975.
[30] GA Nichols, YH Koo, SN Shah, Delay of insulin addition to oral combination therapy
despite inadequate glycemic control: delay of insulin therapy, J. Gen. Intern. Med. 22
(2007) 453458.
[31] JD Lane, CC McCaskill, PG Williams, PI Parekh, MN Feinglos, RS Surwit, Personality
correlates of glycemic control in type 2 diabetes, Diabetes Care 23 (2000)
13211325.
[32] G Nefs, J Speight, F Pouwer, V Pop, M Bot, J Denollet, Type D personality, suboptimal
health behaviors and emotional distress in adults with diabetes: results from Diabetes MILESThe Netherlands, Diabetes Res. Clin. Pract. (2015)http://dx.doi.org/10.
1016/j.diabres.2015.01.015 (pii: S0168-8227(15)00023-6).
[33] M Peyrot, RR Rubin, T Lauritzen, SE Skovlund, FJ Snoek, et al., Resistance to insulin
therapy among patients and providers: results of the cross-national Diabetes Attitudes, Wishes, and Needs (DAWN) study, Diabetes Care 28 (2005) 26732679.
[34] N Ratanawongsa, JC Crosson, D Schillinger, AJ Karter, CK Saha, et al., Getting under
the skin of clinical inertia in insulin initiation: the Translating Research Into Action
for Diabetes (TRIAD) Insulin Starts Project, Diabetes Educ. 38 (2012) 94100.
[35] S Nam, C Chesla, NA Stotts, L Kroon, SL Janson, Factors associated with psychological
insulin resistance in individuals with type 2 diabetes, Diabetes Care 33 (2010)
17471749.
[36] RW Grant, DJ Wexler, AJ Watson, WT Lester, E Cagliero, et al., How doctors choose
medications to treat type 2 diabetes: a national survey of specialists and academic
generalists, Diabetes Care 30 (2007) 14481453.
[37] FJ Snoek, SE Skovlund, F Pouwer, Development and validation of the insulin treatment appraisal scale (ITAS) in patients with type 2 diabetes, Health Qual. Life
Outcomes 5 (2007) 69.